ASSAY
OF
PROTEOLYTIC
ACTIVITY
IN THE
DIAGNOSIS
OF
PANCREATIC
FUNCTION
By Robert G. Frazier, M.D.
Department of Pediatrics, College of Medicine, State University of Iowa
(Accepted August 14, 1958; submitted November 4, 1957.)
PRESENT ADDRESS: 1801 Hinman Avenue, Evanston, Illinois.
332
PEDIATBICS, February 1959
I
N RECENT YEARS increased reliability ofdiagnosis of cystic fibrosis of the
pan-creas has been afforded by the
determina-tion of the concentration of electrolytes in sweat. Several reports have appeared
sug-gesting that normal or partial deficiency of pancreatic function may occur in these
pa-tients, although few documented cases
sup-port this concept.13 These developments give renewed emphasis to the need for
pre-cisc methods of assay of pancreatic
en-zymes and perhaps repeated study of the
course of activity of the enzymes in
duo-denal fluid from these patients.
The assay of proteolytic activity in
duo-denal fluid as a means of evaluating pan-creatic function became commonplace fol-lowing the development of the
Andersen-Early test.4 Whatever its limitations, it is
simple to perform and not time consuming. Recently a modification of the older, more tedious method used by Northrop and
Hussey5 in their studies of trypsin, was
pro-posed by Leubner and Shwachman,#{176} which was purported to be simpler to perform and more accurate. This method assays
proteo-lytic activity by measurement of the rate of
change in viscosity of a gelatin substrate by
means of an Ostwald viscosimeter. Figure 1
is a graph prepared from data published by
the latter authors, which shows the poor
correlation in the results they obtained by
simultaneous studies of duodenal fluid using the two tests. Convincing evidence was not offered to support the contention
that the differences in the results were due to inaccuracies in the Andersen-Early method.
In Figure 1 the general trend of the data would be expected to proceed from lower left to upper right. Both scales are
logarith-mic, which would preserve the expected
straight-line correlation between results ob-tamed by the two tests but make it im-possible to show a zero point indicating
complete absence of activity. The patients
in Leubner and Shwachman’s report with
cystic fibrosis of the pancreas all showed measurable proteolytic activity by both
methods, but there is little evidence of the
same degree of correlation that is seen in the closely grouped data from normal
con-trols. There is also the disturbing
sugges-tion in the symbols linked by arrows,
repre-senting determinations in the same subject
at successive times, that a patient may pro-gress from a state of normal pancreatic ac-tivity to almost complete loss of activity as measured by one test without evidence of change in the other.
The present paper reports studies of some
of the assumptions, kinetics and technical
problems in the performance of these two
tests, and discusses the evaluation of data
from the author’s laboratory and of pub-lished data on proteolytic activity in duo-denal fluid.
PRESENT STUDIES
Materials and Methods
Necessary equipment for the gelatin-viscosity test includes a constant-temperature water bath, in a glass tank, so that viscosimetric read-ings can be easily made. Viscosimeters of the Ostwald type were obtained through the Mac-Alaster-Bicknell Corporation, 243 Broadway, Cambridge, Massachusetts. This type of vis-cosimeter, which can remain clamped to the edge of the water bath, allows the combined solutions of enzyme and substrate to be
in-cubated in the bulb in one arm and to be drawn into the capillary tube in the other arm
for measurement of flow-time at desired
x
CYSTIC
FIBROSIS
OFTHE
PANCREAS
I
a
a
x
.
.x
x
x
*
x
x
x
x
x
x
x
volume
of duodenol
fluid
.16
.08
.04
.02
.01
.005
.0025
ml.
0.1
‘ I 1I
I
I
I
5
10
25
50
100200
ANDERSEN-EARLY
TEST
fr-
x
0.5
Ftc. 1. Comparison of gdatin-viscosity test and Andersen-Early test. Data graphed are taken from work of Leubner and Shwachman.d Arrows join data obtained from
dilution Vo
same patient at different times. Coefficient of activity, K =
-
,x
logtime Vt
20
Cci)
U
a’-ci)
0
U >
>
(_)
>
(I)
0
V.
(I)
>
z
czt
.
NORMAL
solution for the Andersen-Early test was used
throughout.
The proteolytic activity in the duodenal
fluids from 43 patients, obtained under suitable
conditions, was measured by the
gelatin-viscosity test as described by Northrop and
Hussev and by the Andersen-Early test,4
using aliquots from the same specimen. In 11
Of these patients the diagnosis of cystic fibrosis of the pancreas was established by clinical ex-amination, roentgenogram of the chest and analysis of electrolytes in sweat and proteolytic
activity in duodenal fluid. Only two patients of this latter group had sufficient proteolytic activity in the duodenal fluid to measure
ac-curately by both of the methods described.
RESULTS
General Considerations
The viscosity of a gelatin solution that has undergone proteolytic digestion is much less than the viscosity of a similar
solution of undigested gelatin. The
flow-time through a capillary tube of a
viscosi-meter is an index of the viscosity of a
gela-tin solution and can be used as an indirect measure of the concentration of undigested substrate. Accuracy would depend in part
On the presence of a linear relation between control solutions of different concentrations and their respective flow-times.
Data plotted in Figure 2 show the
flow-time of the diluents used in making the gelatin solution, i.e., water and 5% sodium bicarbonate solution, as indicated on the ordinate. The solid line shows the flow-times of solutions of gelatin ranging in con-centration from 0 to 4% as might occur in the gelatin-viscosity test. As can be seen the relationship is approximately linear
140
20
I00’
80
60
40
2
0/ 00/ ZO/ Aol
‘ -r/o
Concentration
of
Gelatin
FIG. 2. Relation of flow-time to concentration of gelatin. Gelatin solution is corn-posed of 1 ml of 5% sodium bicarbonate solution, 1 ml of water and 2 ml of an
aqueous solution of gelatin varying from 0 to 8% in concentration. Flow-times
of 5% 50(liUIll bicarbonate 5()liltion and water are indicated 011 the ordinate for
comparison.
PANCREATIC FUNCTION
changes in the viscosity of the substrate as measured by the flow-time, would lead to
inaccuracy unless a correction factor were
used.
In this test it must also be assumed that
ProductS of digestion of the gelatin do not
affect the viscosity and flow-time of the
un-digested gelatin, because both moieties
re-main in solution. At the concentrations of
substrate at which it is practical to measure
viscosity with the type of viscosimeter used, the initial phase of the reaction is of the first-order type; in other words, it proceeds
at a rate proportional to the concentration
of the substrate. This can be shown by the fact that a graph of the logarithm of the
(V-VH2#{128})
flow-time, or relative viscosity -,
1120
against the length of time during which di-gestion has occurred, will give a straight
line. Northrop and Hussey5 showed that
(J 0
C 0 U
Q) U)
C
a)
E
F-0
LL
this relationship was useful in the
calcula-tion of enzyme activity, though true only for a decrease of about 20% in the viscosity;
after this the reaction apparently slowed
down. This initial straight portion of the
graph and subsequent slowing of the rate
of the reaction are shown in Figure 3. The enzyme does not lose activity in this short
period; other experiments have shown that if used in lesser concentration, digestion of
the substrate proceeds at a constant rate for
a proportionately longer time. The lack of precise correspondence between concentra-tion of gelatin and flow-time shown in
Fig-ure 2 would not affect linearity of rate of
change over this small a range. It seems likely that products of digestion have a
sig-nificant viscosity themselves and have
in-terfered with the measurement of the vis-cosity of the undigested substrate.
de->
-4--(I)
0 U
U,
> a)
>
.4--0
a)
2ml
‘7.5%Gelatin
ImI
5%NoHCO3
1 ml
1/200
Duodenol
juice
.
2
1000
2000
3000
Time
in
Seconds
FIG. 3, Rate of change in relative viscosity of gelatin substrate during proteolysis. Tlw scrnilogarithmic plot shows the relatively straight portion of the curve (fitted by sight) during the initial 20 to :30% drop in viscosity, followed by a gradual
de-Vt - V11,0
celeration of the rate of change. Relative viscosity, is --- where Vt is
- .VH0
the flow-time of the substrate at any time, t, and V11 is the flow-time of water.’
TiIiiC, t, graphed in seconds on the abscissa, is nwasured from the moment the
solutions of enzyme and substrate are combined.
signed to measure initial reaction rates by serial measurements of the flow-time. As modified by Leubner and Shwachman, only readings of an initial flow-time and a final
flow-time after 1 hour of digestion are
recommended. \Vith no knowledge of the
concentration of enzyme in a specimen of (luodenal fluid, such a procedure would
lead to many tests in which either the di-gestion was complete within a few minutes
or, at the other extreme, had not proceeded
far enough in 1 hour to measure accurately. Calculated activity under such conditions
would be erroneous and usually lower, to a
greater or lesser degree, than that more
ac-curately determined from serial
determina-tions giving the initial reaction rate. Data
obtained to illustrate this are shown in Figure 4. Aliquots of substrate were di-gested with varying quantities of a stock
solutioii Of \Tiokasehi* for a standard
diges-0 Dried powder of whole pancreas, VioLin
Cor-poration, \lonticcllo, Ill.
tion-period of 1 hour. The amount of
diges-tion increased with larger amounts of
en-zyme as shown by the increasing ratio of
initial to final flow-times. The graph shows that only at the lowest concentration of enzyme used (where V0/V = 1.4) did the
calculated activity of the enzyme approach
a stable value. This then is the only sample in which the rate of change of viscosity is
in the linear phase; the only sample which roughly measures the initial reaction rate.
There is, however, no way to tell which of
these aliquots measures the true value, if
only one sample is analyzed using a fixed
length of time for the period of incubation and digestion. By doing serial determina-tions of the early changes in flow-times,
concentrations of enzyme varying over a wide range can he measured in one test.
It may be calculated readily that the ratio of V,,/Vt = 1.4 shows a drop of slightly
more than 25% in the final flow-time (V1) as compared to the initial flow-time (V0). It
16
12
>>
10
0
-aco
1.4
1.6
1.8
3.0
PANCREATIC FUNCTION
value for calculated activity of the enzyme would be even more closely approximated
by the use of smaller amounts of enzyme in the experiment described.
As regards the Andersen-Early test, it is
at once evident from inspection of the dilu-tions of duodenal fluid used in this test that the accuracy is related to the number of dilutions used. In this test the end-point is the tube in which the gelatin fails to
solidify on cooling. For the simpler needs
of a clinical diagnostic test a small number
of tubes covering a wide range of proteo-lytic activity would be sufficient, whereas
for the purposes of quantitative determina-tion of degrees of reduction in proteolytic
activity a greater number of dilutions are necessary. It is also evident that difficulties
in reading the end-point may arise
he-cause in some cases gelatin is neither
com-pletely liquefied nor firmly solidified in a given tube, making it difficult to decide which of two tubes is the correct end-point.
Attention to details affecting all enzymatic reactions, such as temperature during the
reaction and the precise time for which the
reaction proceeds, is important to both
methods.
Comparison of Methods
Recognizing these inherent inaccuracies and the difficulties in the two tests, an at-tempt was made to obtain a valid
compari-50fl of the two methods. The gelatin-viscos-ity test was performed in the manner
recommended by Northrop and Hussey, in
14-
8-
6-
4-2-
increasing
digestion
I I I I 1 1 I i
ao
2.2
24
2.6
2.8
RATIO
_Y2.
at various
dilutions
of Viokose
Vt
Fic. 4. Relation of apparent or calculated activity of enzyme to degree of
diges-tion. Range of possible values for activity, K, indicates the value of serial deter-minations of flow-time rather than use of a single determination after a fixed
TABLE I
RESULTS OF ASSAY OF PROTEOLYTIC ArIVITY
IN DUODENAL JUICE
Gel
aim-(Iase A ndersen- . , .
. vlscoslly Diagnosis
o. Early Teal
‘ Test
which serial readings of the flow-time are obtained until the flow-time has decreased
to about 20% below the initial reading.
Figure 5 shows that, when indicated pre-cautions are observed, a fair correlation
exists between the results obtained by the
simultaneous determination of proteolytic activity with these two methods. Only two
of the patients with cystic fibrosis of the
pancreas listed in Table I had sufficient activity measurable to he included on the graph in Figure 5. It is not possible to make
a strict comparison of the data in Figure ‘5 with the data of Leubner and Shwachman
in Figure 1 because of the small number of
patients in Figures 1 and 5 with cystic
fibrosis of the pancreas who show sonic
cvi-dence of partial pancreatic function. On the
basis of this discussion, it is possible to question the superiority of the gelatin-vis-cosity test alleged on the basis of the
changes shown in the results of the two successive tests in Figure 1 in the case of
the t\V() P1tients claimed to show
)rogres-sive loss of pancreatic function. The data
in both Figure 1 and Figure
5
show, in marked contrast, a good correlation of the results of both tests for a large group of patients with miscellaneous conditions.Figure 6 shows the distribution curves
of data from the initial report of Andersen
and Early’ Oil 24 l)atieits and data
accumu-lated in the authors laboratory on 32
pa-tients, relating iii both instances to
pan-creatic function in children with
miscel-laneous conditions other than cystic
fibro-sis of the Iacreas. On the left of the figure
are graphs of the two groups of data
re-Prtl in die ternis of tile units in which
they were measured. On tile right of the
fig-ure the data have been combined and the
distribution curve transfornied into a shape approximating a iiornil distribution curve
by graphing the frequency against the
logarithm of the functioii measured. Use of
this transformation (into a so-called
log-normal distribution) permits the valid use of normal-curve theory for calculation of
the mean and standard deviation. The
values obtained for the mean and standard
1 <3.1r2 .00 1
2 <3.1 .11
3 <3.1t .14
4 <SIt 0.1 I
5 6
7
S
<3.1t
<3 .Ut
<‘3.U
<3.1
0.1
0. 1 Cystic fibrosis of
0.1 1pancreas 0.1 9 10 11 lz 1’3 <3.H 3.1i* 3,12 5 ri.; 0.1 1 .4ti .IS 9.3 ‘3.9
14 r25 6.1
15 25 6.9
16 rt5 4.8 I
17 50 4.5 I
18 50 l3.
19 .50 3.1
0 .5() 4.5 1
;.ii 30 11.3
‘in 50 6.6
23 50 7.6 1
4 50 5,5
25 .50 .5.9 1 Miscellaneous
clis-26 ‘27 ‘28 5() 5() 50 10.5
S.1
7.2
I eases, usually
sug-gestive of cystic
Ii-brosis of pancreas
l9 50 11.8
30 5() 87.0**
31 100 7.4
3i 100 6.8
33 100 14.8
34 lOt) 8.7 I
35 100 6.9
36 lOt) H.7
37 100 .0 I
38 100 8.4 I
39 100 6J2
40 100 18.3
41 4 43 200 100 0O 13.9 9.91,
47..5J
Siblings of patients
with cystic fibrosis
of pancreas
4- 1/dilutionXO.5.
t Activity coefficient (method of Northrop and
Hus-sey1).
*4- Specimen discarded before discrepancy could be
confirmed.
deviation are shown on the graph.
a
NORMAL
x
CYSTIC
FIBROSIS
OF PANCREAS
.
.
to.
(I)
a)
F-
.4--U)
0
U
(I,
>
.
.
.
Fluid,
ml.
.0025
Volume
of
.16
.08
.04
.02
312
Duodenol
.0l
.005
I
I2.5
5
5b
#{243}o
oo
ANDERSE N -EARLY
UNITS:
X0.5
F’i;. s. Comparison of Andersen-Early and gelatin-viscosity tests (author’s (lata).
No patients with complete pancreatic insufficiency arc shown. Trend of data is
similar to that in Figure 1. Only one determination deviates from fairlv CloSe
grouping of other data.
the :32 patients studied by the author have
i)een grouped and graphed in Figure 7.
As can be seen, the distribution curve is
markedly skewed unless subjected to a log-normal transformation.
The frequency distribution of the
proteo-lytic activity in the duodenal juice of
pa-tients with cystic fibrosis of the pancreas has not been graphed because sufficient
data is lacking, both in published reports
and in the patieilts studied in the author’s
laboratory.
Discussion of Results of Present Study
In comparing two methods purporting to
neasure the same function, poor correlation
of results indicates that the t\V() ITlethOdS
do not, in fact, neasure the same function
or for technical reasons one or both are
in-accurate. In theory the two tests used in
the present studs’ vould be expected to give the same results within the limits of
their accuracies. In both, the substrate and
digestive enzymes are the same; the only difference is the manner of measuring the
change in the physical properties of the
gelatin substrate. Careful analyses of pro-teolvtic activity of duodenal fluids in(licate
that, in the author’s experience, the
ANDERSEN-EARLY
TEST
x
- 24 PATIENTSS - 32 PATIENTS
25
5
25
50
00
200
.5
.0
.5
2.0
25
Units
I
Idilution
x .5
Log Units
Fic. 6. Frequency-distribution curves of proteolytic activity in Andersen-Early test. In graph
00 left, (x) denotes patients of Andersen and Early’ and (.) denotes patients studied in author’s
laboratory. Two groups are combined in log-normal curve on right. No patients with cystic
fibrosis of panCreas are inCluded.
simplifying the performance of the test by making it only necessary to read the flow-time at 1 hour, introduces a large potential for error by ignoring the initial rate of the
reaction. the Andersen-Early method and a
pro-posed modification of the gelatin-viscosity iTIethi)d for assay of proteolytic activity.
Data presented herein show that this modi-fled gelatin-viscosity test, though perhaps
8
6
4,
2’
x
0
4 82
6
Activity
Coefficient
Fic. 7. Frequency-distribution curves of data obtained using gelatin-visCosity test. Total number
of patients studied was :32 as in previous figure.
1.0
.5
Log Activity
2.0
(I-) F-z LU
F-(I)
F-LU
F-
Critical use of these tests may be re-quired in the analysis of duodenal fluid from patients with cystic fibrosis of the pancreas in whom one may seek to detect
only partial reduction or progressive loss in the concentration of enzymes in the
duo-denal juice. Likewise, in investigations of
relatives of patients with this disease, more careful quantitation of activity of enzymes will be necessary if such investigations are
to be fruitful. The critique of the gelatin-viscosity and the Andersen-Early tests pre-sented here demonstrates some of the ways,
and the degree, in which these tests may be in error.
ANALYSIS AND INTERPRETATION OF
RESULTS IN THE LITERATURE
In the literature dealing with pancreatic
function in children, there are few reports containing large samples of hi4’
fewer that have presented the data so that
it can be “‘#{176} and none in which
it can be confirmed that the data have been subjected to valid statistical analysis. In
neither of the reports in which data have been presented has the skewed distribution been 910 Despite this, the
appli-cation of standard statistical methods used for analysis of data with normal
distribu-tion has been 24 In such cases,
one standard deviation below the mean often exceeds the range of the data or gives a negative value! Evaluation of pancreatic function in adults has involved similar
diffi-213 The problem of interpreting
data with skewed distributions occurs rela-tively frequently in biologic material and
is exemplified in the results presented in this paper.
J
ustification for the use of the transfor-mation x = log x has been discussed by Gaddum.8 There is no question that all parameters of measurement of a givenpop-ulation of items may not be normally dis-tributed. In such situations, the use of a logarithmic, or other, transformation of the
data will often change the distribution to a more nearly normal one and allow the satis-factory use of normal curve theory and
standard statistical methods of analysis. It is also worthy of emphasis that most of the clinical reports of evaluation of other enzymes in duodenal fluid, or of proteolytic enzymes determined by other methods than those discussed here, have not presented complete data nor has it been analyzed in a fashion such as this.h14 These reports are
therefore of limited value to the clinical investigator.
SUMMARY AND CONCLUSIONS
Difficulties in the proper use of the
gela-tin-viscosity test for assay of proteolytic
activity have been discussed. It is the
an-thor’s opinion that for routine clinical diag-nostic studies the Andersen-Early test is sufficiently accurate and much simpler.
The considerations presented regarding measurement of pancreatic function and interpretation of data justify a critical at-titude toward accepting some of the pub-lished data regarding normal and partial or
progressive loss of pancreatic function. That such possibilities may exist is not de-nied, but the actual incidence must be determined by proper application and in-terpretation of results from sound methods
for assay of pancreatic enzymes in duo-denal fluid.
REFERENCES
1. di Sant’Agnese, P. A. : Fibrocystic disease
of the pancreas with normal or partial
pancreatic function ; current views on
pathogenesis and diagnosis. PEDIATRICS,
15:683, 1955.
2. Gibbs, C. E., Bostick, W. L., and Smith,
P. M. : Incomplete i)ancreatic deficiency in cystic fibrosis of the pancreas.
J.
Pediat., 37:320, 1950.
3. Shwachman, H., et a!.: Cystic fibrosis of
the pancreas with varying degrees of pancreatic insufficiency. Am.
J.
Dis.Child., 92:347, 1956.
4. Andersen, D. H., and Early, M. V.: Method of assaying trpsin suitable for routine use in diagnosis of congenital
pancreatic deficiency. Am.
J.
Dis.Child., 63:891, 1942.
5. Northrop,
J.
H., and Hussey, R. C. : Amethod for the quantitative
ARTICLES
6. Leubner, H. , and Shwachman, H. : A new
method for assay of proteoltic activity of diiodenal fluid compared with other methods in the study of fibrocvstic dis-ease of the pancreas (mucoviscidosis).
PEDIATRICS, 15:135, 1955.
7. Haas, E. : On the mechanism of invasion.
I. Antinvasin I, as an enzyme in plasma.
J.
Biol. Chem., 163:63, 1946.8. Gaddum,
J.
H. : Lognormal distributions. Nature, 156:463, 1945.9. Klumpp, T. C., and Neale, A. V. : The
gastric and duodenal contents of normal
infants and children; the duodenal en-zyme activity and the gastric and
duo-denal reactions (H-ion). Am.
J.
Dis.Child., 40:1215, 1930.
10. V#{233}gheli, P. V. : Pancreatic enzymes; nor-mal output and comparison of different
methods of assay. PEDIATRICS, 3:749,
1949.
1 1. Richmond, R. C., and Shwachman, H.:
Studies of fibrocystic disease of the pancreas (mucoviscidosis); chymo-trvpsin activity of duodenal fluid. PEDI-ATRICS, 16:207, 1955.
12. Lagerlof, H. 0. : Pancreatic Function and
Pancreatic Disease Studied by Means
of Secretin. New York, MacMillan, 1942, p. 165.
1:3. Dreiling, D. A., and Hollander, F. : Studies
in pancreatic function. II. A statistical
study of pancreatic secretion following
secretin in patients without pancreatic disease. Gastroenterology, 15 :620, 1950.
14. Shwachman, H., Leubner, H., and Catzel,
P.: Mucoviscidosis. Advances Pediat., 7:249, 1955.
AN APPRAISAL OF FULMINANT \IENINC.OCOCCEMIA WITH REFERENCE TO THE
SiiwAnrz-MAN PHENOMENON, W. Margaretten et a!. (Am. J. Med., 25:868, December, 1958.)
The object of this study was to evaluate the evidence for and against operation of
the Shwartzman phenomenon ( local or generalized) in meningococcemia. It is based
on 52 cases of nieningococcus infection seen at the Cincinnati Children’s Hospital in a
20 year period. The fatal cases were compared with 152 patients who survived
meningococcus infection. Vascular thrombosis appears to be the primary lesion in-volved in the production of the hemorrhagic lesions of the skin and adrenals in acute meningococceniia. It is suggested that these lesions may be produced by a local
Shwartzman phenomenon.
Of special interest is the occurrence of three instances of renal cortical necrosis, such
as is seen characteristically in the generalized Shwartzman reaction. Because of the similarity of the lesions found in the fatal cases of meningococcus infection with the
Sliwartznian phenomenon, and the fact that cortisone is one of the most efficient
methods of potentiating the Shwartzrnan phenomenon, led the authors to question the wisdom of the use of adrenal steroid therapy in fulminating meningococcus septicemia.
It was also found that of 156 patients with meningococcus infection, not in shock at the tiiie of admission, there was a mortality of 21 for the patients who received
adrenal steroid therapy compared with a 7% mortality in those who did not receive
